Secure Consensus Control of Multiagent Cyber-Physical Systems With Uncertain Nonlinear Models

Achieving consensus over a class of multiagent systems (MASs) under cyberattacks is studied in this article. The existing literature on secure consensus control of under-attack MASs is based on linear properties of agents models, whereas in practice, linearization may not be feasible in the presence of model uncertainties. Based on this motivation, the main contribution of this article is secure consensus control of MASs in the presences of uncertain nonlinearities in agents models. An MAS is considered consisting of a set of normal agents and a set of attacked malicious agents. A criterion is developed under which each normal agent at each time instant selects safer interaction links to avoid divergence from a consensus/agreement state in the presence of the unknown malicious agents. Accordingly, a network of nonlinear robust controllers is proposed such that under the selection criterion and in the presence of uncertain nonlinearities in the agents models, consensus among the normal agents is guaranteed. Numerical examples validate the accuracy of the proposed consensus control scheme.